Evolution and Mechanism of Fatty Acid Synthase Multienzymes
Inauguraldissertation zur
Erlangung der Würde eines Doktors der Philosophie vorgelegt der
Philosophisch-Naturwissenschaftlichen Fakultät der Universität Basel
von
Syed Habib Tahir Bukhari aus Pakistan
Basel, 2015
Originaldokument gespeichert auf dem Dokumentenserver der Universität Basel edoc.unibas.ch
This work is licenced under the agreement
„Attribution Non-Commercial No Derivatives – 3.0 Switzerland“ (CC BY-NC-ND 3.0 CH).
The complete text may be reviewed here:
creativecommons.org/licenses/by-nc-nd/3.0/ch/deed.en
3
Genehmigt von der Philosophisch-Naturwissenschaften Fakultät auf Antrag von
Prof. Dr. Timm Maier Prof. Dr. Roderick Lim Basel, den 23 Juni 2015
!
Prof.Dr. Jörg Schibler Dekan
5
Summary
Fatty acids are central components of biological membranes, serve as energy storage compounds, and act as second messengers or as covalent modifiers governing protein localization. Biosynthesis of fatty acids uses a conserved mechanism across all species and is carried out in repeated cycles of reactions. In Eukaryotes, these reactions are catalyzed by type I fatty acid synthases (FAS), large architecturally diverse, multienzyme complexes that integrate all steps of fatty acid synthesis into complex biosynthetic assemblies. Two strikingly different types of FAS have emerged in fungi and in animals. The fungal FAS is a rigid, 2.6-MDa barrel- shaped structure with its 48 functional domains embedded in a matrix of scaffolding elements, which comprises almost 50% of the total sequence and determines the emergent multienzymes properties of fFAS. All functional core domains of fFAS are derived from monofunctional bacterial enzymes, but the evolutionary origin of the scaffolding elements remains enigmatic. In the first part of the thesis using a combined phylogenetic and structural biology approach we have identified two bacterial protein families of non-canonical fatty acid biosynthesis starter enzymes and trans-acting polyketide enoyl reductases (ER) as potential ancestors of core scaffolding regions in fFAS. The architectures of both protein families are revealed by representative crystal structures of the starter enzyme FabY and DfnA-ER. In both families, a striking structural conservation of insertions to scaffolding elements in fFAS is observed, despite marginal sequence identity. The combined phylogenetic and structural data provide first insights into the evolutionary origins of the complex multienzyme architecture of fFAS.
In contrast structural and evolutionarily analysis revealed that animal FAS is related to polyketide synthase type I (PKS I), which is utilized by bacteria to synthesize a broad spectrum of secondary metabolites. Animal FAS is
7
an open X-shaped structure with catalytic domains not interrupted by the insertion of scaffolding elements but connected to each other via short not conserved linker sequences. Crystallographic data together with biochemical and electron microscopy (EM) analysis indicate that animal FAS displays an extraordinary degree of flexibility to ensure productive interactions between the active sites during the reaction cycle.
Conformational changes most likely result from a combination of internal domain flexibility in the linker regions, which connects individual domains in the animal FAS. The second part of the thesis is thus dedicated to investigating how intra domain linking influences catalytic properties and conformational crosstalk between domains. This was achieved by generating more then 40 different constructs with various linker lenths.
Combined structural and kinetic data from purified constructs helped us to better understand the emergent properties of the megasynthase system. A long-term goal is to use these insights for the construction of artificial multienzymes incorporating complete and complex molecular pathways.
Table of Contents
Summary ... 6
List of Figures ... 9
List of Tables ... 10
List of Abbreviations ... 11
Introduction ... 12
Bacterial FAS ... 14
Mitochondrial FAS ... 19
Animal FAS ... 20
Fungal FAS ... 25
Aim of the thesis ... 30
Part I Evolutionary origins of the multienzyme architecture of giant fungal fatty acid synthase ... 33
Summary ... 35
Introduction ... 36
Materials and Methods ... 39
Results ... 42
Discussion ... 56
References ... 61
Supplementary Information ... 65
Part II Emergent properties of Dynamic Multienzymes: The influence of interdomain linking on animal FAS multienzyme kinetics ... 95
Summary ... 97
Introduction ... 98
Materials and Methods ... 102
Results ... 107
Conclusion ... 110
References: ... 111
Summary and outlook ... 112
Acknowledgements ... 117
References: ... 118
9 List of Figures
Introduction
Figure 1. Activation of ACP by AcpS 12
Figure 2. Biosynthesis of Malonyl-CoA by ACC 13 Figure 3. Catalytic reaction cycle of type II bacterial FAS. 15 Figure 4. Structural overview of mammalian FAS 19 Figure 5. Comparison of the catalytic reaction cycle of type II
bacterial FAS and mammalian FAS 21
Figure 6. Structure of the fungal FAS 24
Figure 7. Location the ACP domains. 25
Figure 8. Activation of fungal ACP. 26
Figure 9. Catalytic reaction cycle of the fungal FAS. 27 Figure 10. Distribution of animal FAS conformations (adopted from
Brignole et al., 2009). 30
Part I
Figure 1. Comparison of bacterial and fungal TIM-barrel-fold ERs. 41 Figure 2. Structure of DfnA-ER and comparison to bacterial and
fungal homologues. 45
Figure 3. Location and interactions of the ER domain in fFAS. 46 Figure 4. Comparison of bacterial and fungal KS proteins and
domains 48
Figure 5. Structural analysis of FabY 51
Figure 6. Interactions of the KS domain in fFAS 56 Figure 7. Origin and development of the fFAS multienzymes
architecture 58
Figure S1 (related to Figure 1). Sequence alignments of
DfnA/FabK/ER 69
Figure S2 (related to Figure 2). Domain organization of the difficidin cluster from B. amyloliquefaciens FZB42 and comparison of FabK and DfnA-ER
77
Figure S3 (related to Figure 2). Comparison of DfnA-ER and the fFAS
ER 78
Figure S4 (related to Figure 4). Sequence alignments of FabY
homologues 79
Figure S5 (related to Figure 5). Comparison of FabF and FabY. 88 Figure S6 (related to Figure 6). DfnA-ER has small overlap with the
interdomain interface between the AT and ER domain in fFAS 89 Figure S7 (related to Figure 7). Schematic comparison of fungal FAS I subtypes and distribution of single and split genes of fFAS in the fungal kingdom
90
Part II
Figure 1. Structural overview of mammalian FAS 106
List of Tables
Table 1. Statistics on diffraction data and refinement of FabY and
DfnA-ER 43
Table S1. Protein sequences used for protein alignment and
phylogenetic analysis 66
Table 1. Primers sequence for the KR-ACP deletion constructs 101 Table 2. Primers sequence for the KR-ACP insertion constructs 101 Table 3. Primers sequence for MAT-DH deletion constructs 102 Table 4. Primers sequence for MAT-DH insertion constructs 102
Table 1. Kinetic analysis of porcine FAS 107
Table 2. Table 2. Summary of the all cloned linker constructs for the
MAT-DH region 107
11 List of Abbreviations
ACC Acetyl-CoA carboxylase ACP Acyl carrier protein
AcPS holoACP-synthase
ATP Adenosine triphosphate C14 14-carbon fatty acid C16 16-carbon fatty acid C18 18-carbon fatty acid C20 20-carbon fatty acid
CMN-FAS Corynebacteria, Mycobacteria, and Nocardia Fatty acid synthase
CoA Coenzyme- A
DfnA Difficidin biosynthesis cluster A
DfnA-ER The enoyl reductase domain of DfnA
DH Dehydratase
E.coli Escherichia coli
EM Electron microscopy
ER Enoylreductase
FAS Fatty acid synthase
fFAS Fungal FAS
FMN Flavin mononucleotide
G3P Glycerol-3-phosphate Hex A/B Hexanoic acid synthase
KR β-ketoreductase
KS β-ketoacyl synthase
LFCA Last fungal common ancestor LPA Lysophosphatidic acid
MAT Malonyl-/acetyl-transferase mFAS mammalian Fatty Acid Synthase MPT Malonyl/palmitoyl transferase domain mtFAS Mitochondrial Fatty Acid Synthase P-PAN 4-phosphopantetheine
PA Phosphatidic acid
pKR Pseudo-ketoreductase
PKS I Polyketide synthase type I pME Pseudo-methyltransferase
PPT Phosphopantetheine transferase
Introduction
In the beginning of the early 20th century it was considered that fatty acids have only two functions- serve as a source of calories and as building blocks for membranes1. In 1929, George and Mildred Burr published two papers, where they demonstrated that fatty acids were an essential dietary constituent2,3. In their experiment they kept rats on strict diet and noticed that if fatty acids were omitted from the food, a deficiency syndrome ensued that often led to death4. After this many other research groups were able to show that fatty acids and their metabolites possess very unique biological roles that is distinctive from its function as a source of energy or as a simple construction unit5-7. A wide range of cellular processes are dependent on fatty acids, from the biosynthesis of essential cellular structural components (membrane phospholipids, lipoproteins, and lipoglycans) and cofactors (lipoate and biotin) to energy storage reserves8-10. Fatty acids participate as components of signal transduction pathways and as docking sites for cytoplasmic signaling proteins such as kinases11. Polyunsaturated fatty acids containing two or more carbon–carbon double-bonds are important as constituents of the phospholipids, where they appear to confer distinctive properties to the membranes, in particular by decreasing their rigidity12.
Storage lipids, such as triacylglycerols, are deposited as fat droplets in large amounts in vertebrate fat cells13. These droplets are surrounded by a protective monolayer of phospholipids and biologically active hydrophobic proteins. Triacylglycerols are released when required by hydrolysis reactions catalyzed by lipases under the influence of hormones14. Subsequent oxidation of triacylglycerols produces more than twice the energy (9 kcal/g) as the oxidation of carbohydrates (4 kcal/g)15.
13
Biosynthesis of fatty acids uses a conserved mechanism across all species and it is carried out in repeated cycles of reactions. In Eukaryotes, these reactions are catalyzed by type I fatty acid synthase (FAS), a large architecturally diverse, multienzyme complexes that integrate all steps of fatty acid synthesis into complex biosynthetic assemblies16,17. In contrast, in dissociated type II FAS system, proteins are all expressed as individual polypeptides from separate genes, these systems are found mostly in bacteria but also in eukaryotic organelles such as mitochondria and plastids18. In the following section I will compare the fundamentally distinct organization of different FAS systems and examine the structural and chemical principals of enzyme reactions.
Bacterial FAS
The biosynthesis of fatty acids is the first step in the formation of membrane lipids and it is essential for all bacterial cells. It involves more than ten separately expressed genes and proteins, which are abundant in the bacterial cytosol. Central to this process is the acyl carrier protein (ACP), a cofactor protein that covalently binds all fatty acyl intermediates19. ACP is one of the most expressed protein in E. coli and is converted to its active holo-form by holoACP-synthase (AcpS) which transfers the 4-phosphopantetheine (P-PAN) prosthetic group from CoA to apo-ACP20,21 (Figure 1). Activated holo-ACP then enters fatty acid biosynthesis cycle, which consist of four stages; Initiation, chain elongation, chain reduction and termination.
Initiation of fatty acid biosynthesis. The first step in fatty acid biosynthesis is the carboxylation of acetyl-CoA by acetyl-CoA carboxylase (ACC) to form the universal extender unit malonyl-CoA (Figure 2). The overall ACC reaction requires a biotin cofactor, adenosine triphosphate (ATP) and the coordinated action of four gene products, AccA, AccB, AccC, and AccD22.
15
Chain elongation. The elongation step is initiated by the Claisen condensation of malonyl- ACP with an acyl-CoA, catalyzed by the condensing enzyme, the ß-ketoacyl-ACP synthase III or FabH, to form ß- ketoacyl-ACP23 (Figure 3 B). FabH of E. coli, produces mainly linear fatty acids, because it has specificity only for acetyl-CoA, in contrast Gram- positive bacteria utilize special FabH enzymes which can choose as a first building block larger branched-chain substrates24.
Chain reduction. The Chain reduction cycle consists of three core enzyme activities that progressively reduce the acyl chain attached to ACP through each round (Figure 3 C). First, the NADPH-dependent ß-ketoacyl- ACP reductase, or FabG, reduces the ß-keto group to a ß-hydroxyl intermediate25. Second, two isoforms, FabA and FabZ, catalyze the dehydration of -hydroxyacyl-ACPs, albeit with different substrate specificities 26. The third step involves the reduction of the enoyl chain by the NADH-dependent FabI27. Gramm-negative bacteria utilize alternative enoyl-ACP reductase — the flavoprotein FabK28. The fully reduced acyl- ACP chain functions as a starter substrate for the next round of
elongation, which is initiated by an elongation condensing enzyme: FabF or FabB (Figure 3C). The FabF isoform is universally expressed, but some bacteria utilize the FabB enzyme, which is used for condensing unsaturated fatty acids26,29.
During each round of the condensation reactions, the acyl chain is detached from ACP and binds to the cysteine residue in the active sites of FabH, FabB or FabF (Figure 3)18. An extender malonyl-ACP then enters the active site and the acyl chain is added to the carboxyl end of the malonyl unit, which loses a CO2 group in the process. Therefore, the acyl chain is constructed ‘inside out’ as the additional carbon groups are added to the base of the acyl chain. The cycle is repeated until the acyl chain reaches 16–18 carbon groups in length, at which point the vast majority of acyl-ACPs are utilized in membrane biosynthesis (Figure 3D)30. Transfer of fatty acids to the membrane. The most ubiquitous system is the PlsX–PlsY pathway, which is found in all but one family of proteobacteria30. First, PlsX, a peripheral membrane protein, transfers the acyl group from the long-chain acyl-ACP end product of the elongation pathway (Figure 3, blue arrow) to inorganic phosphate to form a reactive acyl-phosphate intermediate (Figure 4D). This is then attached to glycerol-3-phosphate (G3P) to form acyl-glycerol-3- phosphate (LPA; lysophosphatidic acid) by the acyltransferase membrane protein PlsY. Another acyltransferase, PlsC then adds a second acyl chain to the 2-position of LPA to form phospatidic acid (PA)31. PA represents the fundamental building block
17
Figure 3. Catalytic reaction cycle of type II bacterial FAS. (A) FabD transfers the malonyl group from CoA to ACP. (B) FabH initiates first cycles of fatty acid elongation by combining acetyl-CoA with malonyl- ACP to form acetoacetyl-ACP. (C) The NADPH-dependent FabG reduces the condensation product to β-hydroxyacyl- ACP. The hydroxyl group is removed by one of two β- hydroxyacyl dehydratases FabZ/FabA. The double bond is then reduced in an NADH-dependent reaction by an enoylreductase FabI generating acyl-ACP extended by two carbon units.
At this point the cycle starts again through the condensation reaction of acyl-ACP with another malonyl-ACP group catalyzed by FabB/F. This is repeated multiple times until saturated C16 or C18 acyl-ACP is diverted for utilization in membrane biosynthesis (blue arrow). (D) The most widely distributed pathway starts with the conversion of a long-chain acyl-ACP end product of fatty acid synthesis to an acyl-P by PlsX. PlsY transfers the fatty acid from the acyl-P to glycerol-3-phosphate G3P to form lysophosphatidic acid (LPA). LPA is then converted to phosphatidic acid (PA) by PlsC. PA is the key intermediate in the synthesis of all membrane glycerolipids.
from which the phospholipids are derived, giving rise to phosphatidylserine, phosphatidylethanolamine and phosphatidyl-glycerol.
Regulation of fatty acid synthesis. The primary pathway for the regulation of fatty acid synthesis in E. coli is through feedback inhibition by long- chain acyl-ACPs, which affects three enzymes: ACC, FabH and FabI.
Inhibition of ACC limits the supply of malonate groups for chain initiation and elongation (Figure 2)32. Regulation of FabH prevents the initiation of new acyl chains and limits the total number of fatty acids that are produced (Figure 3B)33. Finally, FabI catalyzes the reduction of enoyl-ACP, which is critical for the completion of the acyl chain elongation cycle; a reduction in FabI activity slows the rate of fatty acid elongation (Figure 3C)34.
19 Mitochondrial FAS
The mitochondrial FAS (mtFAS) produces short-chain fatty acids, which are essential for the structure, dynamics and enzymatic function of inner and outer mitochondrial membranes35-37. Any irregularity in enzymes, responsible for biosynthesis of mitochondrial fatty acids in eukaryotes results in respiratory incompetence, abnormal morphology and cell deaths38-41.
mtFAS diverges from the cytosolic FAS as it is of the type II dissociated organization and many of the proteins are highly homologous to their bacterial counterparts, but nevertheless the eukaryotic type II systems do have three distinguishing features36. First, the prokaryotic systems utilize three different β-ketoacyl synthases (FabH, FabB and FabF) with divergent substrate specificities that range from 2- to 16-carbon; the mtFAS only has one KS and predominantly produces fatty acids of 14 or less carbon atoms18,42. Second, the bacterial type II system has a FabD, dedicated enzyme which directly utilizes acetyl-CoA in the initiation step43. In contrast, the mitochondrial type II system appears to generate the acetyl primer by decarboxylation of malonyl moieties at the β- ketoacyl synthase35,42. Third, the mtFAS proteins that catalyze the final two steps of the fatty acid elongation cycle, Htd2 and 2-enoyl-ACP reductase (Etr1), do not share clear sequence similarities to prokaryotic FAS type II enzymes and structurally belong to different protein classes42.
Animal FAS
Eukaryotic type I fatty acid synthases (FAS) are giant multifunctional proteins. Various evolutionary processes such as gene duplication and gene fusion led to the emergence of the type I FASs45. In animal cells fatty acid synthesis is catalyzed by a single 540kDa homodimeric multienzyme with a characteristic X-shape (Figure 4)17,46,47. Based on its X-ray crystal structure, mammalian Fatty Acid Synthase (mFAS) is divided into a lower condensing portion containing the β-ketoacyl synthase (KS), malonyl- /acetyl-transferase (MAT) domains and an upper β-carbon modification section, consisting of the enzymatic dehydratase (DH), enoylreductase (ER), and β-ketoreductase (KR) domains17 (Figure 4B and C). The upper part also possesses two additional non-enzymatic domains, a pseudo- ketoreductase (pKR) and a pseudo-methyltransferase (pME), according to their structural homology with active KR and ME enzymes48. All reaction intermediates, like in the bacterial FAS system, are covalently bound to an ACP, which translocates between the active sites during catalysis.
Reaction products are released from the ACP as free fatty acids by the thioesterase (TE) domain (Figure 4D). Interestingly, both ACP and TE domains were found disordered in the crystal structure and could not be visualized due to their flexibility48. Structurally the KS, KR and MAT domains are homologs of their bacterial functional counterparts FabB, FabD and FabG48-50. The DH domain adopts a pseudo-dimeric fold, distantly resembling the bacterial homo-dimeric FabA48.
A key feature distinguishing the type I FASs and type II counterparts is the presence of discrete connecting regions between the active domains. In the porcine FAS 9% of total sequence is invested in
21
the form of solvent exposed linkers48,51. The importances of the linking regions are discussed in part II. Despite differences in the overall
organization of the mFAS, the enzymatic reactions and mechanism of de novo fatty acid synthesis are essentially identical to dissociative bacterial FAS system (Figure 5). For example, exactly the same intermediates and reactions are present in the elongation cycle. However, some enzymatic differences do occur in the stage of initiation, condensation and termination.
The MAT domain transfers acetyl- and malonyl-CoA. The Bacterial FAS system utilizes two dedicated enzymes, FabH and FabB, for transferring acetyl or malonyl-CoA 18,52 (Figure 5B, step 1 and 4). In contrast the animal FAS contain a single MAT enzyme for loading both the Acetyl and Malonyl-CoA units on to the ACP46. The choice of substrate loaded is entirely random acetyl and malonyl moieties are rapidly exchanged between CoA and FAS. If the applicable substrate is loaded, then a productive reaction can follow, otherwise the inappropriately loaded substrate is transferred back to CoA, which must be present at all times during the FAS reaction to ensure efficient substrate sorting4. Scavenging of CoA from the assay incubation mixture halts fatty acid synthesis53,54. TE and KS determine the chain length of fatty acid products. Two enzymatic domains determine the chain length in mFAS. First, during the elongation cycle ACP transfers growing fatty acid chains to the active site cysteine of the KS domain. For fatty acids containing up to 16 carbons this transfer is very rapid and occur in less then 1 second, however for chains containing 18, 20 or 22 C atoms this process requires several minutes55. Secondly, TE has very limited activity toward substrates with less than 16 carbon atoms 56. Thus the specificities of the chain elongation and chain termination steps complement each other perfectly
23
Figure 5. Comparison of the catalytic reaction cycle of type II bacterial FAS and mammalian FAS. Red arrows indicate steps that are repeated only once; black arrow shows reactions that are redone multiple times. Proteins sharing enzymatic functions in animal and type II FAS are colored in identical color. (A) Catalytic cycle of bacterial FAS. Bacteria utilize a specialized enzyme (FabH) for the initiation step in acyl chain formation and uses an acyl-CoA as a primer to condense with malonyl-ACP (step 2 and 3). For more details refer to Figure 3. (B) Catalytic reaction cycle of mammalian FAS. The reaction cycle of FAS is initiated by the transfer of the acyl moiety of the starter substrate acetyl-CoA to the ACP (step 1) catalyzed by the dual-specific malonyl/acetyl transferase (MAT). ACP then transfers the acetyl group to the active cysteine on the KS (step 2). In the next step the elongation unit malonyl-CoA is loaded onto ACP by MAT (step 3 and 4). The ß-ketoacyl synthase (KS) catalyzes the decarboxylative condensation of the acyl intermediate with malonyl-ACP (step 5). The product is further modified at the ß- carbon position by β-ketoreductase (KR) (step 6), dehydratase (DH) (step 7) and enoyl reductase (ER) (step 8) to yield a four carbon acyl substrate for further cyclic elongation with two-carbon units derived from malonyl-CoA (step 4). After seven rounds of elongation, the end product is released from the enzyme as free fatty acid by a thiosterase (TE)
to ensure that the main product released from the FAS is the 16 C atom fatty acid.
25 Fungal FAS
Yeast FAS, a member of the fungal type I FAS family, contains six copies of eight independent functional domains in an α6ß6 molecular complex of 2.6 MDa57 (Figure 6). Each of the α and ß subunits accommodates four functional domains. The ß-chain carries the AT, ER, DH domain, and the largest part of the malonyl/palmitoyl transferase domain (MPT). The remaining half of the MPT, the double-tethered ACP, the KR, the KS and the phosphopantetheine transferase (PPT) are encoded by the α-chain16,58,59 (Figure 6, A and B). These eight functional domains catalyze all reactions required for synthesis of fatty acids in yeast:
activation, priming, multiple cycles of elongation, and termination60. The assembled fFAS adopts a barrel-shaped formation with two domes separated by a central wheel structure16,44,57,58,60 (Figure 6 D). Each dome contains three full sets of enzymatic domains and three double-tethered ACP domains for substrate transfer (Figure 7A)60,61.
Fungal FAS invests nearly 50% of its absolute sequence length into building scaffolding elements, which are mainly inserted of conserved enzymatic domains16,62 (Figure 7B). These sequences are not directly involved in catalysis but instead dictate the architectural interactions and define the arrangement of the catalytic domains 16,58,60. Despite considerable differences in the overall organization of fungal FAS, the enzymatic reactions and mechanism of de novo fatty acid synthesis are essentially identical to dissociative bacterial and X shaped mammalian FAS systems18,50,57. However, some differences occur in the activation of ACP, elongation and termination stages.
27
Figure 7. Location the ACP domains. (A) The ACP domains, shown as red surfaces, are located inside the fungal FAS barrel (golden color). The active site clefts of the enzymatic domains participating in the fatty acid elongation oriented to interact with the ACP. (B) Fungal FAS scaffolding elements (grey) without core enzymatic domains.
Fungal FAS requires a specific activation mechanism. Before the ACP can start to deliver its substrates, it has to be posttranslationally modified by the addition of a P-Pan63,64. This activation of the fungal FAS is performed by a specific PPT very similar to bacterial AcpS that covalently attaches the phosphopantetheine moiety of coenzyme A (CoA) onto a conserved serine residue of the ACP65. One of the major differences of fungal and mammalian FAS is the mechanism of the posttranslational modification of the ACP domain. A separately expressed PPT enzyme performs the activation of the mammalian FAS, in contrast the fungal FAS PPT domain is fused to the C terminal end of the α chain66. This PPT domain is located outside of the barrel, spatially separated from the ACP (Figure 8)16,44,57-60. Therefore it is currently not clear how fungal PPT could activate ACP. One possibility is that the fungal FAS auto-activates during the folding events prior to the closure of the reaction chambers67.
Figure 8. Activation of fungal ACP. Front view of the fungal FAS (right panel) and close- up view of ACP (orange) and PPT (cyan) locations.
Fungal FAS uses a bi-functional MPT for loading and termination. The fFAS harbors no TE domain like in animal FAS, but contains a bi-functional MPT domain instead, which transfers malonyl moieties used for the chain elongation from CoA to ACP and back-transfers saturated C16/C18 products from ACP to CoA58,60,68 (Figure 9). Two factors determine this unique property of the fungal MPT. First MPT contain a deep hydrophobic pocket, which is optimally suited for binding the hydrophobic C16 tail of palmitate58,60. Second during the elongation cycle malonyl- CoA preferentially binds to the active site of MPT. But as soon as growing acyl chain is long enough the binding affinity to the MPT will be strong enough to displace malonyl CoA and allow transferring the mature fatty acid to free CoA57,68-70.
29
Figure 9. Catalytic reaction cycle of the fungal FAS. Fungal FAS utilizes a dedicated AT domain, which is located on ß-chain and has unique specificity for the priming substrate, acetyl-CoA (step 1). In contrast, mammalian FAS use the MAT domain to load both the priming and the elongating substrate onto ACP (Figure 5B, step 1 and 4).
In bacterial type II FAS systems, the acetate primer is directly transferred from acetyl-CoA to the ß -ketoacyl-ACP synthase III (FabH) that catalyzes the first condensation reaction in the chain elongation cycle (Figyre 5A, step 2). Fungal FAS adopts bi-functional MPT for choosing elongation (step 4) substrate and terminating reaction by transferring 18 carbons fatty acid back to CoA.
Aim of the thesis
Nature developed three types of FAS enzymes, built on completely different architectural principles, but catalyzing highly related series of chemical reaction18,57. In the type II FAS, reaction intermediates are covalently attached to the ACP that shuttles substrates between the dissociated enzymatic components18. In the multifunctional eukaryotic FAS, ACP forms an integral part of the catalytic machinery resulting in minimized diffusion distances and higher catalytic efficiency48,60. The 2.6 MDa barrel shaped fungal FAS integrates all active domains in the rigid scaffolding matrix which comprises almost 50% of the total sequence71. Inside the barrel the concentration of ACP and all other active sites is approximately 1 mM ensuring that none of the enzymatic reaction steps are rate-limiting58. Substrate shuttling within the fungal FAS happens entirely by 2D diffusion of the double-tethered ACP, without a requirement for large overall conformational changes61. All functional domains of fFAS are derived from monofunctional bacterial enzymes, but the evolutionary origin of the scaffolding elements remains enigmatic. The first part of the thesis is therefore focused on finding out the evolutionary origins of scaffolding elements using combined phylogenetic and structural biology approach to better understand the evolutionary process, which led to the development of the fungal FAS.
Structural and evolutionarily analysis revealed that animal FAS is related to polyketide synthase type I (PKS I), which is utilized by bacteria to synthesize a broad spectrum of secondary metabolites 50,72,73. Animal FAS is an open X-shaped structure with catalytic domains connected to each other via short not conserved linker sequences17,48. Crystallographic data together with biochemical and EM analysis indicate that animal FAS displays an extraordinary degree of flexibility to ensure productive interactions between the ACP and the active sites during the reaction cycle (Figure
31
10)17,48,50,51. The nature and dynamic aspects of the substrate shuttling mechanisms in animal FAS are not entirely understood. The second part of the thesis is thus dedicated to investigating how inter-domain linking influences catalytic properties and conformational crosstalk between domains. This will be done by generating more then three dozens of different constructs with systematically increasing or decreasing linker lengths in different areas of animal FAS. Combined structural and kinetic data from purified constructs will help us to better understand the emergent properties of the megasynthase system. A long-term goal is to use these insights for the construction of artificial multienzymes incorporating complete and complex molecular pathways.
Figure 10. Distribution of animal FAS conformations (adopted from Brignole et al., 2009).
(a-d) Class average of single particle images (black and white) with calculated three- dimensional structures (yellow). (e) Cartoon of different FAS arrangements in upper and lower part, red (asymmetric) and blue (symmetric) in the upper ß-carbon-processing, faded color represent perpendicular or in plane conformations of the lower part FAS. (f) Representation % of particles in different conformations, bars are colored according to conformations in (e)
33 Part I
Evolutionary origins of the multienzyme architecture of giant fungal fatty acid synthase
Habib S.T. Bukhari, Roman P. Jakob and Timm Maier
Author contributions
Timm Maier, Roman P. Jakob and I designed the project. Roman P.
Jakob produced DfnA protein, obtained crystals and carried out crystallographic analyses.
I established purification, obtained crystals and determined crystal structure of FabY.
I performed bioinformatic analysis of FabY and DfnA. Timm Maier supervised the project, contributed to crystallographic work and analyzed the results.
Roman P. Jakob, I and Timm Maier wrote and revised manuscript.
Manuscript was published in journal Structure (cell press), volume 22, number 12, December 2, 2014
35 Summary
Fungal fatty acid synthase (fFAS) is a key paradigm for the evolution of complex multienzymes. Its 48 functional domains are embedded in a matrix of scaffolding elements, which comprises almost 50% of the total sequence and determines the emergent multienzymes properties of fFAS.
All functional domains of fFAS are derived from monofunctional bacterial enzymes, but the evolutionary origin of the scaffolding elements remains enigmatic. Here, we identify two bacterial protein families of non- canonical fatty acid biosynthesis starter enzymes and trans-acting polyketide enoyl reductases (ER) as potential ancestors of core scaffolding regions in fFAS. The architectures of both protein families are revealed by representative crystal structures of the starter enzyme FabY and DfnA-ER. In both families, a striking structural conservation of insertions to scaffolding elements in fFAS is observed, despite marginal sequence identity. The combined phylogenetic and structural data provide first insights into the evolutionary origins of the complex multienzyme architecture of fFAS.
Introduction
Fatty acids are central components of biological membranes, serve as energy storage compounds, and act as second messengers or as covalent modifiers governing protein localization. In most eukaryotes, their biosynthesis is catalyzed by giant multifunctional enzymes, the fatty acid synthases (Type I FASs) 46,57,74, while bacteria and plants employ a series of monofunctional enzymes (Type II FAS) 18,75,76. All FAS systems are built upon a conserved set of chemical reactions and enzymatic activities: An acetyl primer and malonyl elongation substrates are loaded from coenzyme A (CoA) to the phosphopantetheinylated acyl carrier protein (ACP) by acetyl- and malonyl-transferases (AT and MT) and are condensed to acetoacetyl-ACP under decarboxylation by ketoacyl synthase (KS). In three subsequent reaction steps, the β-carbon group is processed by ketoacyl reductase (KR), dehydratase (DH), and enoyl-reductase (ER) to yield a saturated acyl-ACP elongated by a two-carbon unit. This product serves as a primer for the next round of elongation and the elongation cycle continues until a chain length of C16 or C18 is reached. In a terminating step, fatty acids are back-transferred to CoA or released by thioesterase (TE). The eukaryotic Type I FAS integrates all these enzymatic activities required for de novo fatty acid biosynthesis into unique protein assemblies catalysing more than 40 reaction steps. These Type I FASs are prototypic paradigms for the general trend in eukaryotes towards the formation of larger multidomain proteins, which minimize unspecific interactions and permit advanced regulation of localization, activity and degradation 44.
Two strikingly distinct Type I FAS have evolved in eukaryotes, the metazoan and the fungal FAS (fFAS). The metazoan FAS is a 540 kDa homodimer with two complete sets of functional domains and a versatile architecture defined by a minimal amount of scaffolding elements 17,48. Structural
37
analysis of metazoan FAS and bacterial polyketide synthases (PKS) revealed a common architecture 50,72,73, which is used in PKS to synthesize a broad spectrum of secondary metabolites 44,77. A fully methylating, iterative bacterial PKS was identified as a common evolutionary ancestor of metazoan FAS and modular PKS based on the presence of an evolutionary remnant of a methyltransferase domain in the metazoan FAS structure 45,48.
Fungal FAS forms a 2.6-megadalton assembly comprising 48 functional domains, as exemplified by the yeast α6β6-heterododecameric FAS. In addition to the enzymatic activities for fatty acid elongation, it may also incorporate a phosphopantetheinyl transferase (PPT) domain, for cofactor attachment to the ACP. All domains are embedded into a scaffolding matrix that comprises nearly 50 % of the total mass and mediates the majority of architectural interactions determining the spatial arrangement of catalytic centres 16,58,60-62. fFAS adopts a unique barrel- shape structure with two domes enclosing two reaction chambers, each housing three sets of functional domains, separated by a central wheel structure (Figure 1A). This architecture is shared with the more recently described, closely related CMN-FAS systems in Corynebacteria, Mycobacteria, and Nocardia 74,78, which have a slightly lower number of scaffolding expansions and lack an internal PPT 79,80. The CMN- and fFAS are amongst the most complex biosynthetic protein machineries known 62. Still, the evolutionary appearance of the hallmark scaffolding matrix for integrating functional domains, which defines the architecture of fFAS (and CMN-FAS) remains enigmatic, as no intermediate steps of assembly formation have been identified so far.
Extension to core conserved folds are notoriously difficult targets for the analysis of homology and phylogeny as well as for structure prediction, because overall sequence conservation in these regions is
extremely weak and strictly conserved motifs, e.g. representing catalytic sites, are absent. Thus, we use a hybrid approach of bioinformatic analysis guided by and combined with experimental structure determination as a gold standard for the analysis of the evolution of the fFAS scaffolding matrix 81,82. Bioinformatically, we identify potential evolutionary ancestors of fFAS by searching for homologues of fFAS domains that carry insertions to their core folds in equivalent positions as their fFAS relatives. Crystal structures of candidate proteins reveal their structural organization and unambiguously demonstrate the fFAS-like organization of the respective insertion elements.
39 Materials and Methods
Sequence data retrieval, alignment and phylogenetic analyses
The amino acid sequences of all proteins were retrieved from GenBank (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi). A BlastP search was performed using the protein sequence of FabK, fFAS ER, FabF and fFAS KS as the query sequence against completed bacterial and fungal genomes. A total of 47 ER domains and 80 KS domains derived from the complete genome survey were subjected to a phylogenetic analysis (Table S1). Sequences from each enzyme family were selected to have 40-80 % sequence identity to each other. Alignments were created using ClustalW and adjusted manually based on structural alignments using Geneious version 6.0. Rooted phylogenetic trees were generated by the Neighbor-Joining method using a Jukes-Cantor distance model.
Bootstrapping was done using 100,000 random seeds, which were replicated 10,000 times.
Cloning, expression and purification of DfnA-ER
The enoyl reductase domain of DfnA (A7Z6E3, res. 300-752) from Bacillus amyloliquefaciens FZB42 (DSM 23117) 83,84 was cloned from genomic DNA and cloned into the expression plasmid pNIC28-Bsa4 85. Here the protein is linked to a N-terminally hexa-His-tag followed by a TEV-protease cleavage site. The protein was overproduced in E. coli BL21(DE3) pRIL pL1SL2 86. Cells were lysed by sonication in 50 mM Hepes/NaOH, 500 mM NaCl, pH 7.4, 20 mM imidazol and the supernatant was cleared by centrifugation. DfnA-ER was purified by immobilized metal-affinity chromatography on a Ni-NTA column (elution with 250 mM imidazole), with His-Tagged TEV-protease digested 87, followed by a Ni-NTA column step and then subjected to size- exclusion chromatography in 20 mM Hepes/NaOH pH 7.4, 250 mM NaCl, 5 % Glycerol and 5 mM DTT on a Superdex S200 column (GE Healthcare).
The protein-containing fractions were pooled and concentrated in Amicon Ultra units (Millipore).
Cloning, expression and purification of FabY
FabY (PA5174) was PCR-amplified from Pseudomonas aeruginosa PAO1 genomic DNA and cloned into the expression plasmid pNIC28-Bsa4 85. The protein was overproduced in E. coli BL21(DE3) pRIL pL1SL2 86. Cells were lysed by sonication in 50 mM Hepes/NaOH, 500 mM NaCl, pH 7.4, 40 mM imidazol and the supernatant cleared by centrifugation, FabY was purified by metal-affinity chromatography on a Ni-NTA column (elution with 250 mM imidazole), and then subjected to size-exclusion chromatography in 20 mM Hepes/NaOH pH 7.4, 250 mM NaCl, 5 % Glycerol and 5 mM DTT on a Superdex S200 column (GE Healthcare). The protein-containing fractions were pooled and concentrated in Amicon Ultra units (Millipore).
Protein crystallization and structure determination
FabY crystals grew in sitting drop setups at 4 °C at a protein concentration of 8-12 mg/ml using 0.2 M Li2SO4, 0.1 M Bis Tris pH 5.5, and 15 % polyethylene glycol 3350. Crystals were flash frozen in liquid nitrogen after addition ethylene glycol to 25 % (v/v). DfnA-ER crystals were obtained at room temperature in 15 % PEG3350, 0.1 M sodium malonate, 0.1M Bis Tris at pH 6. Crystals were flash frozen after gradually increasing the ethylene glycol concentration to 20 % (v/v) over 2h. Data were collected at beamlines PXI and PXIII of the Swiss Light Source (Paul Scherrer Institut, Villigen, Switzerland) and processed using XDS 88,89. FabY crystals belong to space group C2221 with unit cell parameters of a= 99.4 Å, b = 123.3 Å and c= 100.6 Å and two molecules per asymmetric unit. Structure determination was performed by molecular replacement with the FabF
41
crystal structure (PDB ID: 1KAS) 90. The final FabY model includes all residues. DfnA-ER crystallized in space group P212121 with cell dimensions a = 80.8 Å, b = 94.0 Å, and c = 144.5 Å and two molecules per asymmetric unit. The final model comprises residues 304 to 751. Residues 300-303, 496- 509 and 752 could not be build in the electron density map. Structure determination was performed by molecular replacement with FabK ER as a search model (PDB ID: 2Z6I) 28. Model building and structure refinement were performed for both structures with Coot 91, PHENIX 92 and Buster-TNT
93 (Table 1).
Data deposition
The atomic coordinates for FabY and DfnA-ER have been deposited in the RCSB Protein Data Bank under the accession code 4cw4 and 4cw5.
Supplemental Information
Supplemental information includes one table and 10 figures
Acknowledgments
We thank the staff at the Swiss Light Source (Villigen, Switzerland) for outstanding support for crystallographic data collection, Prof. Peter Leadlay for providing pL1SL2 and Tina Jaeger for the genomic DNA of Pseudomonas aeruginosa PAO1. This work was supported by the Swiss National Science Foundation R’Equip and Project Funding grants 3106030_145023 and 31003A_138262, respectively.
Results
We hypothesized that scaffolding elements defining the multienzyme structure of fFAS might already occur in bacterial proteins not involved in the formation of assemblies with fFAS-like complexity. To test this hypothesis we focussed on two distinct functional domains of fFAS, the KS and ER. The KS domain is the defining unit of fatty acid synthases responsible for the decarboxylative condensation reaction. Together with the KR, it forms the fFAS α-chain central wheel (Figure 1A). The ER domain of fFAS is located in the central region of the β-chain, which forms the capping domes (Figure 1A). Direct interactions between insertion elements of KS and ER connect the dome and central wheel regions and determine the overall organization of fFAS.
Identification of extended TIM-barrel ERs in trans-AT PKS and PUFA synthases
ER domains display an unusual diversity among FAS systems 57,94: The canonical bacterial ER and the metazoan FAS ER are NADPH-dependent Rossmann-fold enzymes, but the ER of fFAS is a ~550 aa domain comprising a TIM-barrel with a permanently bound flavin mononucleotide (FMN) and a large α-helical insertion. Its only distant structural neighbor is the non-canonical bacterial ER FabK 28, a 320 aa dimeric protein, which contains a conserved TIM-barrel but lacks all extension to the barrel observed in fFAS ER (Figure 1B).
43
Fig. 1. Comparison of bacterial and fungal TIM-barrel-fold ERs.
(A) Crystal structure of yeast FAS (PDB ID: 2UV8). The KS, ER and KR domains are colored in orange, green and yellow respectively. (B) Sequence organization of the ER domains of FabK, trans-AT PKS, PUFA, CMN-FAS and fFAS, at approximate sequence scale.
CMN-FAS, fFAS, trans-AT PKS and PUFA specific sequences are colored in orange and purple, respectively. A PUFA-ER specific N-terminal extension is shown in green. Insertions present in trans-AT PKS, CMN- FAS and fFAS are shown in yellow. The same color code is used throughout. (C) Phylogenetic tree and distribution of TIM barrel ER proteins among bacteria and fungi. The tree is drawn to scale, with branch lengths in the units of number of amino acid substitutions per site. The five main groups FabK, trans-AT PKS, PUFA, CMN-FAS and fFAS are indicated.
We have used the minimal TIM-barrel ER FabK for the identification of TIM- barrel enoyl reductases in sequence similarity searches and reconstructed a phylogenetic tree for this family using a neighbor-joining algorithm (Figure 1B and 1C). This analysis identifies two families of ER domains that are more closely related to fFAS ER than FabK: ER from trans-AT PKS and the PfaD family of ERs in marine polyunsaturated fatty acid synthases (PUFA). PfaD homologues are standalone enzymes 95-97 and share a distinct ~80 aa N-terminal extension, unique to PUFA ERs (Figure 1B;
green). Trans-AT PKS ER domains may occur as trans-acting isolated proteins (e.g. PedB, EtnA), but in most PKS systems they are attached to one or two AT domains to form a trans-acting AT-ER protein 98. Members of both protein families have not been characterized structurally, but with 450 to 480 residues they are about 40 % larger than FabK (Figure 1B and Figure S1); the average sequence identities between trans-AT PKS ER and PUFA ER domains with fFAS ER are ~15 %. Sequence alignments identify three insertion sites in trans-AT PKS/PUFA ER as compared to FabK: Two adjacent insertions (ER-S1/S2) are specific for trans-AT PKS/PUFA ER domains (Figure 1B; magenta), whereas a large insertion (Figure 1B;
orange) (ER-S3) overlaps with a major insertion element in fFAS ER (Figure S1).
The trans-AT PKS DfnA is a dimeric FMN-dependent enoyl reductase
To reveal the structural organization of PUFA- and trans-AT PKS ER, we crystallized the representative ER domain of DfnA (aa. 300-752), a trans- acting AT-ER protein involved in difficidin biosynthesis (Figure S2) 83,84. The crystal structure was solved by molecular replacement using FabK (PDB ID:
2Z6I) 28 as search model and refined to Rwork/Rfree of 20.7/23.3 % at a resolution of 2.3 Å (Table 1)
Table 1. Statistics on diffraction data and refinement of FabY and DfnA-ER
45
FabY DfnA-ER
Wavelength (Å) 0.99997 1.0003
Resolution range (Å) 49.7 – 1.35 (1.40 -
1.35)* 47 - 2.30 (2.38 - 2.30)*
Space group C 2 2 21 P 21 21 21
Unit cell 99.4 123.3 100.6 80.9 94.0 144.5
α, β, γ (°) 90 90 90 90 90 90
Total reflections 875960 (78609) 350174 (33349) Unique reflections 134985 (13224) 52945 (5162)
Multiplicity 6.5 (5.9) 6.6 (6.5)
Completeness (%) 99.85 (98.80) 99.93 (99.71)
Mean I/sigma(I) 17.30 (1.59) 17.99 (1.38)
Wilson B-factor 12.97 46.72
R-merge 0.060 (1.005) 0.071 (1.256)
R-meas 0.065 0.077
CC1/2 0.999 (0.60) 0.999 (0.54)
CC* 1 (0.87) 1 (0.84)
R-work 0.148 (0.281) 0.207 (0.311)
R-free 0.177 (0.301) 0.233 (0.333)
Number of atoms 5934 7122
macromolecules 5067 6802
ligands 6 62
water 861 258
Protein residues 638 875
RMS(bonds) 0.009 0.007
RMS(angles) 1.28 1.00
Ramachandran
favored (%) 97.0 98.0
Ramachandran
outliers (%) 0.15 0.23
Clashscore 4.25 1.74
• Values in parentheses are for highest resolution shell.
• Table 1. Statistics on diffraction data and refinement of FabY and DfnA-ER
with two virtually identical molecules per asymmetric unit. The DfnA-ER monomer consists of a (β/α)8 TIM-barrel domain (aa. 305-603 and 701-752) with a bound FMN cofactor and an inserted α-helical substrate-binding domain (aa. 604-700) (Figure 2A). The (β/α)8 TIM-barrel domain closely resembles FabK with an r.m.s.d. of 1.7 Å over 321 matching Cα atoms.
DfnA-ER dimerizes via a large 1920 Å2 interface formed by its TIM-barrel domain and an extension of a C-terminal α-helix (dimerization tip) as observed for FabK. The structural analysis reveals that the ER-S1/S2 insertions expand the TIM-barrel domain relative to FabK by forming a small subdomain opposite to the dimerization interface and located 30 Å away from the active site (Figure 2A, Figure S3): Insertion ER-S1 forms a β- hairpin (aa. 432-442) which pack against the two helices (aa. 463-488) of insertion ER-S2.
The major insertion in DfnA is structurally conserved in fungal FAS
DfnA-ER exhibits very low sequence conservation (12.7 % identity) to the ER domain (Figure S1) of the fFAS β-chain (aa. 583-1109 in yeast FAS).
Nevertheless, the DfnA-ER crystal structure reveals a close relationsship to the fFAS counterpart with an r.m.s.d. of 1.97 Å over 360 matching Cα atoms (Figure 2B) 99. Short insertions in the TIM barrel domain relative to the FabK core fold are either specific to DfnA (insertions ER-S1, ER-S2, see above) or fFAS (aa. 784-794, 827-840), where they are involved in contacts to the neighboring AT domain (Figure 3).
The larger ER-S3 (aa. 628-674) insertion of DfnA-ER uses exactly the same insertion site as the major extension of CMN- and fFAS-ER, the C1 insertion 61 comprising residues 879-1024 in yeast FAS (Figure 2B). The expansion regions aa. 628-650 and 651-674 in DfnA match in a structural superimposition aa. 879-904 and 997-1024 in yeast FAS (Figure S4).
47
Fig. 2. Structure of DfnA-ER and comparison to bacterial and fungal homologues. (A) Cartoon representation of DfnA-ER. The FMN cofactor is shown in green, the dimerization tips of the C-terminal helix are shown in red, extensions ER-S1/S2 (magenta) and ER-S3 (yellow) are indicated. Anchor points for the N-terminally attached AT domain in full- length DfnA are indicated in pink. (B) Extension of the ER core fold of FabK in trans-AT PKS, CMN-FAS and fFAS. Structures of FabK (PDB ID: 2Z6I), DfnA-ER and fFAS ER from S.
cerevesiae (PDB ID: 2UVA) and CMN ER of S. smegmatis (PDB ID: 3ZEN) are shown in cartoon representation.
Fig. 3. Location and interactions of the ER domain in fFAS. Cartoon representation of S.
cerevesiae FAS (PDB ID: 2UV8) in front (top) and top view (bottom). The core of the ER domain is colored in dark grey, ER expansion segments present in trans-AT PKS, CMN-FAS and fFAS are shown in yellow and FAS specific helical insertions are colored orange. The AT domain is colored in pale pink. The extension of the KS core fold are highlighted DM3/butterfly (green), DM3/shoulders (cyan), CIS/arms (red) and DM4 (dark blue). Close- up views of the inter-subunit interactions mediated by ER (β-chain) and KS (α-chain) in yeast FAS are shown on the right side.
49
651-674 in DfnA match in a structural superimposition aa. 879-904 and 997- 1024 in yeast FAS (Figure S4). Inbetween these conserved segments, a further five-helix bundle (aa. 905-996) is inserted in fFAS. As evidenced by their absence from FabK, the ER-S3 insertion is not required for a general stabilization of the core fold or a productive active site conformation. In DfnA-ER it also has apparently no relevance at the level of the isolated ER- domain: It is neither involved in dimerization nor in DfnA-specific adaptations of the active site. Its conservation specifically in trans-acting (AT)xER proteins and PUFA rather suggests an involvement in the formation of interdomain or transient intermolecular interactions in PKS assembly lines, an analogy to the role of intersubunit connection C1 for bridging α- and β-subunits in fFAS (Figure 3).
FabY is the closest monofunctional relative of the CMN- and fFAS ketosynthase
The fungal FAS KS domain with a length of ~720 residues is much larger than its monofunctional bacterial counterparts FabH (~ 320 aa) 52 and FabB/F (~400 - 420 aa) 29. This is mainly due to three large insertions (Figure 4): the dimerization modules 3 (DM3; green and cyan) and 4 (DM4; dark blue) and a C-terminal insertion (CIS; red) 61.
DM3 forms a core part of the central wheel and is involved in ACP binding of KS 60,61. DM4 is located at the periphery of the KS dimer and provides the attachment point for the PPT domain, wheras CIS is involved in interactions with the ER domain in fFAS (Figure 3).
Fig. 4. Comparison of bacterial and fungal KS proteins and domains. (top) Linear sequence organization of the KS proteins FabF and FabY and the KS domains of CMN-FAS and fFAS, at approximate sequence scale. The DM3 domain is colored cyan and green, the DM4 and CIS insertions are shown in blue and red, respectively. (bottom) Phylogenetic tree and distribution of KS I/II domain-containing proteins among bacteria and fungi.
The tree is drawn to scale, with branch lengths in the units of number of amino acid substitutions per site.
51
In a bioinformatic search we identified proteins of the FabY family as the most extended homologues of the complete fFAS KS domain (Figure 4).
FabY acts as the starter enzyme for fatty acid biosynthesis in P. aeruginosa by catalyzing the condensation of acetyl moieties from acetyl-CoA to malonyl-ACP. Its deletion affects growth, siderophore secretion, quorum- sensing signaling and lipopolysaccharide synthesis 100,101. Members of this recently described family of ketosynthases 102 are ~ 630 residues in size, about 200 aa larger than other monofunctional bacterial KSs from the FabB/F family (Figure 4). Our phylogenetic analysis shows that FabY shares a common evolutionary ancestor with the KS domain of CMN- and fFAS (Figure 4). FabY has only 19 % overall sequence identity to fFAS KS, but the fFAS insertions DM3 and CIS have correspondences in FabY (Figure S5).
FabY lacks the DM4 insertion of fFAS, which is involved in PPT attachment.
Interestingly, DM4 is also absent in bacterial CMN-FAS, which utilize trans- acting PPTs instead of integrated ones, as exemplified by mycobacterial FAS 79,80. As a result, FabY is strikingly similar over its full length to the KS domain of mycobacterial type I FAS (26 % identity over 600 residues).
Large core-fold extensions in the non-canonical P. aeruginosa starter KS FabY
To analyze the similarity of fFAS KS and FabY at a structural level, we crystallized the dimeric 140 kDa P. aeruginosa FabY yielding crystals in space group C2221, with one monomer per asymmetric unit and the dimeric assembly generated by crystallographic twofold-symmetry. The structure was solved by molecular replacement using FabF (PDB ID:
2GFW) and refined to Rwork/Rfree 14.8/17.7 % at 1.35 Å resolution. FabY is a member of the α/β-hydrolase superfamily (Figure 5A) 103,104. Structural superimpositions of the FabY core fold only with those of the three
bacterial ketosynthase families reveals a close match (for 370 aligned residues) to both FabB (r.m.s.d. 1.6 Å) 105 and FabF (r.m.s.d. 1.8 Å) 90 (Figure S6), whereas FabH is structurally more distantly related (r.m.s.d. 2.6 Å). In line with the core fold similarity, FabY uses a Cys-His-His (Cys281, His434 and His474) triad typical for elongating KSs (FabB/F) enzymes (Figure 5B), whereas the functional orthologs of FabY, the FabH starter condensing enzyme, is characterized by a His-Asn-Cys catalytic triad (Yuan, Sachdeva et al. 2012) (Figure 5B). In the high resolution crystal structure, alternate conformations are observed for the active site cysteine residue (Cys281), which are also detected in the spatially adjacent loop 532-535. The active site residues of FabY have similar orientations as in the elongating KS homologues FabB/F (Figure 5B), whereas the acyl pocket is much shorter and resembles the starter condensing enzyme FabH (Figure 5B). This is consistent with the finding that FabY utilizes only short chain acyl-CoA as substrates 101.
While the bacterial KSs FabB/F closely resemble the catalytic core of FabY, its overall closest relative are the KS domains of CMN- and fFAS, to which the entire FabY superimposes with an r.m.s.d. of 1.9 Å and 2.0 Å over 570 matching residues, respectively 95. In comparison to the bacterial canonical ketosynthases, FabY has three noticeable expansion segments, Shoulder, Arms and Butterfly (Figure 5). The Shoulder region (aa. 33-83) is inserted close to the N-terminus and comprises two α-helices and a small three-stranded β-sheet laterally positioned away from the two-fold symmetry axis of dimeric FabY. The Butterfly (aa. 98-158), follows only ten residues later and consists of a shaft
53
Fig. 5. Structural analysis of FabY. (A) Cartoon representation of dimeric (dark and light colours) FabY (B) Active site of FabH (top) (PDB ID: 1NHJ) 100, FabY (middle) and FabF (bottom) (PDB ID: 2GFY)101. The substrate entry and acyl pocket of the KS domains are oriented to the left and right, respectively. Active site residues and the dodecanoic acid (red) bound to FabF are shown in stick representation. Dual conformations of the active site Cys281 in FabY are indicated. The acyl pocket in the starter KS FabH and FabY are significant shorter than in the elongation KS FabF. (C) Extension of the KS core fold of FabF in FabY, CMN- and fFAS. Structures of FabF (PDB ID: 1GFW), FabY and the KS domains of S. cerevesiae (PDB ID: 2UV9) and M. smegmatis (PDB ID: 3ZEN) FAS are shown in ribbon representation. The insertion of butterfly (green), shoulders (cyan) and arm (red) in FabY are structurally conserved in CMN- and fFAS. The fFAS specific DM4 is shown in dark blue.
formed by two two-stranded anti-parallel β-sheets, capped by another three-stranded β-sheet. The Arms are formed by the 80 C-terminal residues of FabY (aa. 548-635). They originate from the terminal β-strand of the core fold and comprise a short and a very long helix spanning the height of the core domain. The helices are followed by an extensive linker (aa.
586-606) without regular secondary structure elements that protrude all across to the second protomer. The terminal region of the Arms forms a small three-stranded antiparallel β-sheet, before ending in a long loop (Figure S7).
All expansion elements are located on the periphery of FabY distant from the active site. While the Shoulders do not contribute to dimerization, both the Arms and the Butterfly expansion contribute considerably to the overall dimerization interface of FabY via contacts to the core fold or the synonomous expansion regions of the second protomer, respectively.
The three expansion elements strikingly resemble the structure of insertion elements observed in fungal and mycobacterial FAS multienzymes (Figure 5C). The Arms closely match the CIS insertion comprising residues 3012-3089 and 1659-1711, respectively, in CMN- and yeast FAS. The Butterfly and Shoulder region together resemble an insertion element designated as DM3 in fungal (aa. 1118-1179) 16 and mycobacterial FAS (aa. 2553-2615) 79. DM3 in fFAS is a component of the central wheel structure and provides part of the binding interface for the ACP-KS interaction 60,61. Based on sequence analysis and the conserved connecting region in between (Figure 4 and Figure S5), we suggest that the two expansions, Butterfly and Shoulder, resulted from independent insertion events and may have separate functions. The functional relevance of the insertion regions remains to be uncovered: Butterfly and Arms contribute to dimerization, however, an equivalent extended
55
dimerization interface is not required in other bacterial ketosynthases with a conserved dimeric structure.
